This case study outlines the events of a recent summative computer-aided assessment (CAA) failure involving 280 first-year undergraduate students. Post-test analysis found that the central server had become unexpectedly overloaded, thereby causing the CAA to be abandoned. Practical advice on just what to do in the event of a summative CAA failure is virtually non-existent in the related literature. In response, this paper provides a detailed account of the post-failure analysis and recovery activities from a practitioner's (ie, academic) viewpoint. Supported by empirical evidence, the research shows how the concept of "optional substitution" was developed as a pragmatic, equitable, and broadly acceptable solution to the problem.